Qwen 2.5 Max: benchmark results
Alibaba's API-only MoE flagship pretrained on 20T+ tokens, launched to rival DeepSeek V3 and GPT-4o (January 2025). Provider: Alibaba. Released 2025-01-28. Access: API.
Aggregate Rating 1607 ± 12, rank #416 of 1656 rated models, from 143 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass Language - Dialogue - Chinese (CompassBench 2501) | 83.7 | Score (%) | 100 |
| OpenCompass Language - NLP - English (CompassBench 2501) | 82.8 | Score (%) | 100 |
| OpenCompass LLM - Language (CompassBench 2501) | 75.5 | Score (%) | 97.4 |
| OpenCompass LLM - Language - Chinese (CompassBench 2501) | 74.7 | Score (%) | 97.4 |
| OpenCompass LLM - Language - English (CompassBench 2501) | 76.2 | Score (%) | 97.4 |
| OpenCompass Language - Creation - Chinese (CompassBench 2501) | 76.5 | Score (%) | 97.4 |
| OpenCompass Language - Creation - English (CompassBench 2501) | 74.6 | Score (%) | 97.4 |
| MERA - SimpleAr | 100 | EM (%) | 95.5 |
| OpenCompass Language - Dialogue - English (CompassBench 2501) | 71.2 | Score (%) | 94.9 |
| MERA - ruTiE | 91.07 | Accuracy (%) | 94.8 |
| OpenCompass Code - Execution - English (CompassBench 2501) | 29.1 | Score (%) | 93.6 |
| OpenCompass Knowledge - Science - English (CompassBench 2501) | 89 | Score (%) | 93.6 |
Interactive version: theaggregate.ai/model?slug=qwen-2-5-max · How It Works · Data refreshed daily, snapshot 2026-10-20.